On Thu, Apr 01, 2021 at 12:34:04AM CDT, Andrew Jeffery wrote:

On Thu, 1 Apr 2021, at 15:48, Zev Weiss wrote:

quoted

On Wed, Mar 31, 2021 at 11:15:44PM CDT, Joel Stanley wrote:

quoted

On Thu, 1 Apr 2021 at 00:57, Zev Weiss [off-list ref] wrote:

quoted

This provides a simple boolean to use instead of the deprecated
aspeed,sirq-polarity-sense property.

Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
---
 drivers/tty/serial/8250/8250_aspeed_vuart.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/tty/serial/8250/8250_aspeed_vuart.c b/drivers/tty/serial/8250/8250_aspeed_vuart.c
index c33e02cbde93..e5ef9f957f9a 100644
--- a/drivers/tty/serial/8250/8250_aspeed_vuart.c
+++ b/drivers/tty/serial/8250/8250_aspeed_vuart.c

@@ -482,6 +482,9 @@ static int aspeed_vuart_probe(struct platform_device *pdev)
                of_node_put(sirq_polarity_sense_args.np);
        }

+       if (of_property_read_bool(np, "aspeed,sirq-active-high"))
+               aspeed_vuart_set_sirq_polarity(vuart, 1);

This assumes the default is always low, so we don't need a property to
set it to that state?

Would it make more sense to have the property describe if it's high or
low? (I'm happy for the answer to be "no", as we've gotten by for the
past few years without it).

Yeah, that sounds like better way to approach it -- I think I'll
rearrange as Andrew suggested in
https://lore.kernel.org/openbmc/d66753ee-7db2-41e5-9fe5-762b1ab678bc at www.fastmail.com/

quoted

This brings up another point. We already have the sysfs file for
setting the lpc address, from userspace. In OpenBMC land this can be
set with obmc-console-client (/etc/obmc-console.conf). Should we add
support to that application for setting the irq polarity too, and do
away with device tree descriptions?

I guess I might lean slightly toward keeping the DT description so that
if for whatever reason obmc-console-server flakes out and doesn't start
you're better positioned to try banging on /dev/ttyS* manually if you're
desperate.  Though I suppose that in turn might imply that I'm arguing
for adding DT properties for lpc_address and sirq too,

Why not just adopt exactly what I've done with KCS, where we have aspeed,lpc-io-reg and aspeed,lpc-interrupts?

Andrew

Ah -- yes, that does sound like a sensible approach.  I'll send a v3 
with that worked in.


Zev